Solid tumor size for prediction of recurrence in large and giant non-functioning pituitary adenomas

A subset of large non-functioning pituitary adenomas (lNFPA) and giant non-functioning pituitary adenomas (gNFPA) undergoes early progression/recurrence (P/R) after surgery. This study revealed the clinical and image predictors of P/R in lNFPA and gNFPA, with emphasis on solid tumor size. This retrospective study investigated the preoperative MR imaging features for the prediction of P/R in lNFPA (> 3 cm) and gNFPA (> 4 cm). Only the patients with a complete preoperative brain MRI and undergone postoperative MRI follow-ups for more than 1 year were included. From November 2010 to December 2020, a total of 34 patients diagnosed with lNFPA and gNFPA were included (median follow-up time 47.6 months) in this study. A total of twenty-three (23/34, 67.6%) patients had P/R, and the median time to P/R is 25.2 months. Solid tumor diameter (STD), solid tumor volume (STV), and extent of resection are associated with P/R (p < 0.05). Multivariate analysis showed large STV is a risk factor for P/R (p < 0.05) with a hazard ratio of 30.79. The cutoff points of STD and STV for prediction of P/R are 26 mm and 7.6 cm3, with AUCs of 0.78 and 0.79 respectively. Kaplan–Meier analysis of tumor P/R trends showed that patients with larger STD and STV exhibited shorter progression-free survival (p < 0.05). For lNFPA and gNFPA, preoperative STD and STV are significant predictors of P/R. The results offer objective and valuable information for treatment planning in this subgroup. Supplementary Information The online version contains supplementary material available at 10.1007/s10143-021-01662-7.


Introduction
Pituitary adenomas (PA) constitute 10-25% of all intracranial neoplasms [12]. A subgroup of these tumors particularly challenging to manage are those that can be classified as large and giant PA [4,8]. Although there is no consensus regarding the exact definition of tumor size, the largest tumor diameter of > 4 cm is considered giant, whereas > 3 cm is considered large [5,8,19,30,34,38]. Large and giant PA that grow beyond the sellar are always difficult to manage surgically because of the surrounding important neurovascular structures and a greater risk of complications [4,8,35]. Large and giant PA comprise about 6-10% of all pituitary tumors [12,44]. Most of them are clinically non-functioning pituitary adenomas (NFPA) and occur predominantly in males [12,17,44]. Visual field defects resulting from compression of optic chiasm are the most common preoperative symptoms followed by a headache. Partial or total hypopituitarism is observed in some patients due to tumor compression of the normal pituitary gland [17]. Although more than 90% of NFPA are diagnosed as benign adenomas according to 2017 WHO classification [27], 12-46% of them may undergo early progression/recurrence (P/R) after surgical resection [6,7,9,13,40]. Gross-total resection (GTR) by transsphenoidal approach (TSA) is the standard surgical treatment in the majority of NFPA; however, it is difficult to achieve in large NFPA (lNFPA) and giant NFPA (gNFPA) [36]. Therefore, a relatively high P/R rate due to postoperative residual tumor had been reported in this subgroup [6,7,39]. Although adjuvant radiotherapy (RT) is implemented in some institutions to prevent P/R in NFPA, 20-30% of patients may have irreversible hypopituitarism or other complications after treatment [17].
MRI findings such as cavernous sinus invasion, absence of apoplexy, postoperative residual tumor, diffusion restriction, and radiomics score have been reported as significant parameters related to P/R in NFPA [21,28,46]. However, some of them are qualitative, and others need to be analyzed on advanced MRI sequences. In oncologic imaging, Response Evaluation Criteria in Solid Tumors (RECIST), based on simple one-dimensional morphologic measurement of tumor diameter, is the gold standard for assessment of treatment response in solid tumors [10,43]. However, a modified RECIST was developed based on the concept that a viable tumor should be defined as only intratumoral enhancing solid mass in some tumors [24,26]. Although large tumor size is associated with lesser extent of tumor resection and more surgical complications in NFPA [2,13,28,36,41], the preoperative quantitative tumor size for the prediction of postoperative recurrence in NFPA was rarely mentioned. Further, no reports regarding the concept of simple measurements of solid tumor size for the prediction of clinical outcomes in NFPA have been published as of yet. This study evaluated the preoperative clinical and MR imaging characteristics for the prediction of P/R in lNFPA and gNFPA, with emphasis on solid tumor diameter (STD) and solid tumor volume (STV).

Ethics statement
This study was approved by the Institutional Review Board (IRB no. 10902-009). Written consent was waived because the retrospective nature of this study does not influence the health care of the included patients. All patients' medical records and imaging are anonymized and de-identified prior to analysis.

Patient selection
The inclusion criteria of this study are patients diagnosed with large (> 3 cm) or giant (> 4 cm) NFPA by brain MRI and pathological confirmation, and with post-operative follow-up MRIs (at least 2 times) more than 1 year after treatment. Patients with clinical, biochemical, or histopathological evidence of hormone hypersecretion are excluded. Diagnosis of prolactinoma is considered unlikely if the prolactin levels are below 100 mg/L according to previous studies [2,16], and a conclusion thereafter confirmed by immunocytochemical studies. Patients receiving postoperative adjuvant RT before P/R are also excluded. From September 2010 to December 2020, 292 patients are diagnosed with PA in our institution. Thirtyfour patients (34/292, 11.6%) (21 men, 13 women, age 20-80 years; median age, 49.5 years) diagnosed with lNFPA and gNFPA are included in this study by the abovementioned inclusion and exclusion criteria. Among them, thirty-two patients underwent surgery performed by TSA, and 2 patients received both TSA and craniotomy due to large tumor sizes. Fourteen (14/34, 41.2%) patients received repeated surgery due to tumor recurrence. The median follow-up duration for all patients is 47.6 months (range from 12 to 115 months). In 23 patients with P/R, the median time to P/R is 25.2 months (range from 6 to 67 months). Clinical and biochemical data are obtained from medical records.

Extent of resection and progression/recurrence
The extent of resection (EOR) is determined by reviewing postoperative MRI by a neuroradiologist (C.C.K.) and a neurosurgeon (S.W.L.). According to published literatures [21,45,46], GTR is defined as a lesion with a residual tumor volume of less than 10% of its original size. In contrast, subtotal resection (STR) is defined as the presence of a residual tumor more than 10% of its original volume. For determining P/R in the included NFPA patients, pretreatment and postoperative MR images are evaluated by two experienced neuroradiologists (C.C.K. with 7 years of experience and T.Y.C. with 20 years of experience), both of whom are blinded to the clinical outcomes of the studied population. P/R is defined as tumor recurrence after GTR or enlargement of residual tumor after STR based on postoperative contrast-enhanced (CE) T1WI. According to published literatures [2,21,46], the threshold of P/R is defined as a more than 2-mm increase in size of residual tumor in at least one dimension when compared with postoperative MRI studies. Inter-observer reliability in the determination of P/R is obtained via a Cohen k value of 0.9. Judgment is made via consensus for equivocal cases. The preoperative MRI findings, including cavernous sinus invasion (Knosp classification) [20], extrasellar extension (Hardy's classification) [15], compression of the optic chiasm and the third ventricle, hydrocephalus, and intratumoral apoplexy or cystic change are determined on coronal T2WI and CE T1WI.

Imaging acquisition
Preoperative brain MRI images are acquired with a 1.5-T (Siemens, MAGNETOM Avanto) (n = 18), 1.5-T (GE Healthcare, Signa HDxt) (n = 10), or a 3-T (GE Healthcare, Discovery MR750) (n = 6) MRI scanner equipped with an 8-channel head coil in each machine. Scanning protocols include axial and sagittal spin-echo T1-weighted imaging (T1WI), axial and coronal fast spin-echo T2-weighted imaging (T2WI), axial fluid-attenuated inversion recovery, axial T2*-weighted gradient-recalled echo, and coronal and sagittal contrastenhanced (CE) T1WI with fat saturation. Dynamic CE T1WI with a small field of view on the pituitary gland is also performed. For CE imaging, intravenous administration of 0.1 mmol/kg of body weight of gadobutrol or gadoterate meglumine is performed. Detailed imaging parameters in the MRI scanners are described in supplementary file 1.

Measurement of tumor diameter and volume
Measurements of both tumor diameter and volume were obtained on coronal CE T1WI ( Fig. 1) by using the freehand region of interest (ROI) tool on the Picture Archiving and Communication System (PACS) (INFINITT PACS; INFINITT Healthcare, Seoul, Republic of Korea) workstations. The preoperative maximal tumor diameter (MTD) is determined on coronal CE T1WI, and the STD is obtained by the only measurement of the solid tumor part. Preoperative total tumor volume (TTV) and postoperative residual volume (RV) are determined by manually calculating whole tumor areas in each coronal CE T1WI slice, and then compiling the volumes in the z-dimension using a semiautomated PACS measurement tool. The preoperative STV is obtained by the only measurement of solid tumor part without the involvement of intratumoral apoplexy, necrosis, or cystic change, which can be identified on coronal T2WI and CE T1WI (Figs. 1-3).

Statistical analysis
Statistical analyses were performed using SPSS for Windows (V.24.0, IBM, Chicago, IL, USA). For evaluating clinical parameters and MR imaging features, chi-square (or Fisher exact test) and Mann-Whitney U tests are performed for categorical and continuous data respectively. The receiver operating characteristic (ROC) curve of STD, MTD, STV, and TTV for prediction of P/R was performed, and sensitivity, specificity, area under ROC (AUC), and optimal cutoff value were obtained. Further, Kaplan-Meier analyses based on cutoff values of STD and STV were used to evaluate the progression/ recurrence-free survival (PFS), and the log-rank test was used to assess the significance. Cox proportional hazard regression model with univariate and multivariate analysis was performed to determine independent factors of P/R. Variables with a p value < 0.05 in univariate analysis were brought forward to the multivariate analysis. For all statistical analyses, p values < 0.05 were considered statistically significant.
The inter-observer reliabilities in the categorical and continuous data were determined using the Cohen k coefficient and intraclass correlation coefficient (ICC), respectively. The Cohen k coefficient and ICC were interpreted using the methods described by Landis et al. [22]. Both Cohen k coefficient and ICC with values between 0.8 and 1 were obtained, indicating almost perfect agreement. Due to almost perfect reproducibility in the ICC, the subsequent statistical evaluation of continuous data was performed using the mean value calculated from both readers.

Clinical data and MRI features
The clinical data and MRI findings are summarized in Table 1. P/R is diagnosed in 23 (23/34, 67.6%) patients. Large preoperative STD and STV are more frequently observed in the P/R group (p < 0.05) (Figs. 2 and 3). In univariate Cox proportional hazards analysis (Table 2), significantly larger STD/STV and lesser EOR were observed in the P/R group (p < 0.05). Further, large STV is a risk factor for P/R (p < 0.05) with a hazard ratio of 30.79 in multivariate analysis (Table 2).

ROC and Kaplan-Meier analyses in solid tumor size
The median follow-up duration for all patients was 47.6 months. In 23 patients with P/R, the median time to P/R is 25.2 months. The sensitivity, specificity, AUC, and optimal cutoff points of the STD and STV for differentiation between the P/R and non-P/R groups are summarized in Table 3. The cutoff points for the STD and STV ratio were 26 mm and 7.6 cm 3 , respectively. An AUC of 0.78, 0.61, 0.79, and 0.60 were obtained for the STD, MTD, STV, and TTV respectively (Fig. 4). When comparing the tumor P/R trends in Kaplan-Meier analysis, patients with larger STD (more than the cutoff value of 26 mm) and larger STV (more than the cutoff value of 7.6 cm 3 ) exhibited shorter PFS (p < 0.05) (Fig. 5).

Discussion
The purpose of this study was to analyze preoperative solid tumor size in predicting P/R of lNFPA and gNFPA after resection. The results showed that large STD and STV are significantly associated with P/R. Although the risk factors for the recurrence of NFPA have been previously reported, the present results are the first to offer quantitative cutoff points of preoperative solid tumor size for the prediction of P/R in lNFPA and gNFPA. Although most NFPA are benign pituitary adenomas, 12-46% of them may show early P/R within 5 years after surgical resection [6,7,40]. According to 2017 WHO classification [27], tumor invasion and tumor proliferation index (Ki-67 and mitotic count) are associated with aggressive clinical behavior in NFPA. However, the definition of tumor invasion for NFPA was not clear in the WHO criteria and hence cannot be estimated if no corresponding information from MRI studies is considered [40,44]. In the past 30 years, several meta-analyses consistently reported higher recurrence rates in NFPA than in secreting PA [6,33,37,40]. Roelfsema et al. [40] showed that postoperative hormone concentration is an important predictor for P/R in functioning PA; however, no specific predictor is identified for NFPA. In contrast to functioning PA, for which biochemical markers often suggest tumor recurrence before the visible tumor is detected on imaging, tumor remission or recurrence in NFPA is mainly determined by MR imaging [29]. Some MR imaging features for the prediction of EOR and clinical outcomes in lNFPA and gNFPA had been reported [17,18,36]. Invasion of the cavernous sinus, maximum tumor diameter, and absence of tumor apoplexy were associated with an unfavorable surgical outcome in NFPA [28]. Invasion of the cavernous sinus is significantly associated with incomplete resection and residual tumor [2]. In contrast, more complete resection and less tumor recurrence could be achieved in NFPA with apoplexy [1]. This may explain why tumor recurrence is not significantly associated with the largest tumor diameter and total tumor volume, which are measured on both solid and apoplexy/cystic components of NFPA. RECIST is based on one-dimensional measurement of tumor size and is the gold standard for the evaluation of therapeutic response in solid tumors. However, a modified RECIST for hepatocellular carcinoma was developed by Lencioni et al. [26] based on the concept that a viable tumor should be defined as only intratumoral    enhancing solid tumor part [3]. Lee et al. [24] reported that measurement of only solid tumor mass offers a better assessment of treatment response as compared with conventional RECIST in patients receiving targeted therapies for lung cancer. Similarly, association between solid tumor size and recurrence rate in lNFPA and gNFPA were observed in our study. On the other hand, low apparent diffusion coefficient value (diffusion restriction) is associated with recurrence in NFPA [21,42]. Recently, quantitative MRI-based radiomics is also used for the evaluation of tumor behaviors in NFPA [46]. However, these parameters need to be measured on advanced MRI sequences and analyzed with complex statistical algorithms. The concept of simple and straight morphologic measurement focusing on solid tumor size for the prediction of clinical outcomes in NFPA is first mentioned in our study. The results offer quantitative, fast, and consistent measurement for neurosurgeons, radiologists, and clinical physicians in the evaluation of lNFPA and gNFPA. EOR is a significant determining factor in the rate of recurrence in NFPA [2,6,29]. As the residual tumor due to incomplete surgical resection exists in most large and giant NFPA after surgery [18,36], the issue of tumor recurrence is particularly important in this subgroup. Lee et al. [25] reported a recurrence rate of 8.2% and 58.3% in patients receiving GTR and STR, respectively. Maletkovic et al. [29] revealed that postoperative residual tumor confers a tenfold increased risk of recurrence in NFPA. Similarly, the association between EOR and P/R was observed in univariate regression analysis in our study. No statistical difference in the multivariate analysis may be explained by the small sample size and the association between EOR and solid tumor size.
It is known that postoperative adjuvant RT or stereotactic radiosurgery (SRS) is highly effective in preventing P/R in PA [23,31]. Lee et al. [23] showed that empirical SRS was superior to progression-guided SRS for NFPA after subtotal resection. Although adjuvant RT and SRS may increase risks of radiation-induced complications such as hypopituitarism, neurocognitive dysfunction, cerebrovascular disease, and secondary brain tumors, the overall rate of serious complications is low [11,14,17,32]. Progressive and irreversible hypopituitarism is the most commonly reported late complication, up to 20-30% at 5 years after treatment [17]. Since most NFPA are benign tumors, prediction of tumor recurrence offers clinically valuable information for treatment options. For patients with high possibilities of P/R, aggressive surgical resection combined with postoperative adjuvant RT and close MRI follow-up should be considered. In contrast, the surgery would aim to relieve clinical symptoms by decreasing tumor mass effects for patients with lower possibilities of tumor recurrence. Patients receiving adjuvant RT before P/R were excluded from our study because RT may affect the independent prediction of the preoperative MR imaging analysis for P/R.
The results of the current study propose cutoff values of solid tumor size for the preoperative prediction of P/R in NFPA. However, there are still several limitations in this study. Selection bias may exist due to its retrospective nature. As in other ROI-based studies, subjective freehand ROIs might influence the accuracy of the tumor size measurements. The small sample size may limit statistical power to detect potential associations between clinical or imaging parameters and P/R. Finally, there is a lack of Conclusions lNFPA and gNFPA with larger solid tumor part were associated with higher possibilities of recurrence. The preoperative solid tumor diameter and volume for the prediction of P/R offer clinically useful information for the planning of NFPA treatment, including the extent of surgical resection, implementation of post-operative adjuvant RT, and the MR imaging follow-up strategy.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.